Sentry.io
Sentry is how we know when production is misbehaving—our error and release radar for live applications.
If you are not technical: think of Sentry as a smoke alarm for software. When something breaks for real users, it should show up here with enough context to fix it without a log-file treasure hunt.
What Sentry is for (today)
Section titled “What Sentry is for (today)”- Catch errors in staging, UAT, and production before customers report them blindly
- Tie failures to a release (which deploy / which commit introduced the problem)
- Separate environments so staging noise does not look like a production fire
- Support go-live decisions — no Sentry wired correctly → not ready for “fully in production”
Sentry sits alongside Grafana Cloud (Loki logs + Prometheus metrics from Fly). Logs and charts tell the story over time; Sentry highlights what broke and for whom. We plan to cross-link alerting between the two where it helps on-call—details on the Grafana page.
What Sentry is going to do (developing)
Section titled “What Sentry is going to do (developing)”We are standardizing Sentry so it becomes the base for smarter monitoring, not just a dashboard engineers open manually:
| Direction | Plain English |
|---|---|
| Release health | See whether a new deploy made things worse within minutes, not days |
| Source maps | Stack traces point at real code, not minified gibberish |
| AI-assisted triage | Tools that watch, group, and surface recurring server issues—less ad hoc log diving |
| Go-live gate | Part of the Fly checklist sign-off for production |
Exact AI workflows and alerting rules are still being defined. This page will grow as those standards land.
Go-live minimum (summary)
Before we call an API fully in production:
- GitHub:
SENTRY_AUTH_TOKEN,SENTRY_ORG,SENTRY_PROJECT— [Fly checklist](/fly-checklist/) → GitHub Actions information - Fly:
SENTRY_ORG,SENTRY_PROJECT,SENTRY_ENVIRONMENT(andSENTRY_DSNwhere the app reports at runtime) — Fly environment variables panel on the same checklist - Verified: A test error or release visible in Sentry for that environment
Who does what
Section titled “Who does what”| Layer | Who sets it up | What it does |
|---|---|---|
| GitHub secrets | Repo admin / platform | Lets the deploy pipeline upload source maps and tag the release |
| Fly secrets | Platform / on-call per app | Tells the running app which Sentry project and environment to use |
| Application code | Engineering team | SDK sends errors and performance signals |
Patterns vary slightly by repo; new APIs should match the Fly checklist panels rather than inventing a fourth layout.
Related
Section titled “Related”- Grafana — Loki, Prometheus, dashboards (moving off AWS observability VMs)
- Fly.io — health, metrics, org shippers into Grafana Cloud
- GitHub — deploy workflows on sacred branches
- Fly checklist — staging → UAT → production promotion